Time Magazine named ‘You’ the person of the year, officially jumped the shark, and became obsolete all at the same time.
Alright that might have been a bit harsh.
Google patent search is a hot topic right now. Now its easy discover that someone has thought of everything you have, and they have patented it too.
Google also released an sdk. I haven’t looked at it yet, but I will speculate wildly and carelessly.
Without reading a word from google, or really studying the issue, I can say that I think that they released something that allows you to write software in a reasonable way and then ‘compile’ into this very strange pile of web techologies…its some kind of fancy web development framework.
Without reading a single word from google I am going to drop 3 reasons why this is a bad idea.
1.) Building on a mess further solidifies a mess.
Rather then fixing the underlying garbage it builds more infrastructure on top of it.
2.) Abstraction leak. (see joelonsoftware for more details on that this is…)
A modern computer system is an onion. It consists of many many many layers. The innner layers are hardware, drivers, operating system calls, the upper layers are applications and human interfaces. Anytime you work seriously with one of these layers it will leak and the underlying layer will show its ugly head. If you don’t understand the lower level, it will always come back to haunt you. I have found that when seriuosly working iwth computers you cannot neglict a layer simply because there is a pretty layer on top of it. Each layer needs to be as elegant as possible. Each layer requres maintance, costs, troubles, and modifications. Web technologies that currently exists are like English. It just kinda came about because of a need. Its good for getting the point a crossed, but I wouldn’t want to try to explain it to a computer.
3.) Temptation to further obfuscate the html/css/etc layer
It reminds me of trying to automate a gui os by scraping the screen. It’s just backwards in some way.
What’s the solution?
Well, first I should actually read what was released. Can you say shoot first ask questions later?
The bottom line is that people have been writing software for decades now. There was some consensus on how to write an event driven application. Then the web came along andeveryone forgot. A huge step backwards has been taken. This is demonstrated when everyone gets excited when they see a relatively unimpressive application that has existed for 15-20 years (spreadsheet, wordprocessor) running in a web browsers.